Make a multidimensional numpy array

Question

I need to remove the DC part of a signal. To do this I have this code:

def removeDC(seq):
    mean_seq = np.mean(seq)
    seq_mod = np.array([])
    for sample in seq:
        seq_mod = np.append(seq_mod,sample - mean_seq)
    return seq_mod

But my data has the dimensions/shape of (31250, 5), and i want it to remove DC form every channel. This is my try to do this, but im not sure how i add the correct value to the correct channel index

def removeDC(seq):  #removing dc from signal, slik at amplituden varierer rundt 0
    for i in range(0,seq.shape[1])):

        mean_seq = np.mean(seq[:,i])
        seq_mod = np.array([]) #need seq.shape[1] dimensions, problem here
        for sample in seq[:,i]: #problem here
            seq_mod[,i] = np.append(seq_mod, sample - mean_seq)#problem here
    return seq_mod

LudvigH · Accepted Answer · 2021-02-09 14:19:12Z

Another way to frame your problem is like this: You have data in 2 dimensions. Lets call the first one 'row' and the second 'columns' (this is standard vocabulary, but in your application they are probably 'time' and 'channel'). You have 31250 rows and 5 columns. You want to compute the mean for each column, giving you a vector of 5 values. Then, for each of the 31250 rows, you want to remove the 5 mean values. Lets do this!

data = np.random.random(size=(31250,5)) # just some random values...
means = data.mean(axis=0) #compute mean along the columns 
data = data - means

The last line works since numpy identifies your 5 columns with the length 5 of the vector means.

In your specific case, you might want to call them something like

signals  = np.random.random(size=(31250,5))
dc_offset = signals.mean(axis=0)
centered_signals = signals  - dc_offset

PS check out the concept of broadcasting. it is key for writing manageble numpy code. See https://numpy.org/doc/stable/user/basics.broadcasting.html

Collectives™ on Stack Overflow

Make a multidimensional numpy array

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related